ControlPoint Manual — versioned with the code; every worked example below is a REAL corpus job and its numbers are pinned by the test suite.
Staff spikes
What it does
One observer spikes the nail, another rests the flat staff base on it another day — a constant offset (typically 100 mm) between two determinations of the same mark. Every leg has a Spike mm field (default 0, any value); applied offsets are folded into the height difference and LISTED in the report's audit block. When two determinations of a section disagree by ≈ the configured spike height, the software flags STAFF SPIKE SUSPECTED with a one-click chip that applies the reconciling offset to the determination you choose — the software points, the surveyor decides.
Worked example — B01RT01–BM6
Perturbing one determination of the B01RT01–BM6 section by +100 mm (the pinned regression test) fires the detector naming both bookings by their fieldbook:line keys; one click applies −100 mm to the perturbed leg, the network re-adjusts clean, and the offset appears in the audit block with its note.
Theory in one paragraph
In a 1D network, two marks having equal height proves nothing (heights are the unknowns being solved). What identifies a spike is DISAGREEMENT between two determinations of the same physical mark by a suspiciously specific amount — that is a statement about observations, not coordinates, so it survives adjustment and can be tested honestly.